Automated Waste Sorting with Delta Arm and YOLOv8 Detection

Abstract

In the midst of rapid urbanization and industrialization, accurate and efficient waste classification has become an essential task due to the increased emphasis on environmental preservation. Several issues arise from the lack of efficient waste management, such as contamination of the air and water and the spread of disease. Developing nations often face challenges due to limited resources and infrastructure, highlighting the need for effective waste separation. Recent advancements in robotics and machine learning have significantly impacted the waste management sector. This study integrates a robotic arm for effective waste sorting with the most recent version of the You Only Look Once (YOLO) concept, known as YOLOv8. The waste is separated into four categories: paper, plastic, metal, and biodegradable. Inverse kinematics is applied to determine the joint angles needed for the robotic arm to reach a desired position. The results demonstrate that YOLOv8 outperforms state-of-the-art algorithms in waste detection and classification with better precision, recall, and F1 score, emphasizing its potential as a useful tool for enhancing waste management procedures.